Machine Learning Model Selection For Forecasting Entity Energy Usage

Abstract

Embodiments relate to generating time-series energy usage forecast predictions for energy consuming entities. Machine learning model(s) can be trained to forecast energy usage for different energy consuming entities. For example, a local coffee shop location and a large grocery store location are both considered retail locations, however their energy usage over days or weeks may differ significantly. Embodiments organize energy consuming entities into different entity segments and store trained machine learning models that forecast energy usage for each of these individual entity segments. For example, a given machine learning model that corresponds to a given entity segment can be trained using energy usage data for entities that match the given entity segment. A forecast manager can generate a forecast prediction for an energy consuming entity by matching the entity to a given entity segment and generating the forecast prediction using the entity segment's trained machine learning model.

We claim: 
     
         1 . A method for generating time-series energy usage forecast predictions for energy consuming entities, the method comprising:
 storing a plurality of trained machine learning models, wherein each machine learning model is trained to forecast time-series energy usage data for one of a plurality of entity segments;   receiving input data comprising historical time-series energy usage data for an energy consuming entity;   selecting one of the trained machine learning models for the energy consuming entity, wherein the selected trained machine learning model corresponds to an entity segment that matches the energy consuming entity; and   generating a forecast prediction using the selected trained machine learning model and the input data, wherein the forecast prediction comprises a time-series energy usage forecast for the energy consuming entity over a period of time.   
     
     
         2 . The method of  claim 1 , wherein the entity segments are organized according to an organizational structure that comprises a plurality of enterprise type categories, and the organizational structure comprises hierarchical structures under at least a portion of the enterprise type categories. 
     
     
         3 . The method of  claim 2 , wherein each entity segment corresponds to an enterprise type category or a leaf-node of the organizational structures. 
     
     
         4 . The method of  claim 3 , wherein the hierarchical structures under each of the portion of the enterprise type categories are organized according to one or more enterprise parameters with respect to each enterprise type category. 
     
     
         5 . The method of  claim 4  wherein the enterprise parameters comprise one or more of enterprise sub-types, enterprise size, or goods or services for sale. 
     
     
         6 . The method of  claim 4 , wherein selecting one of the trained machine learning models for the energy consuming entity further comprises:
 selecting an enterprise type category that matches attributes of the energy consuming entity, wherein,
 when the selected enterprise type category corresponds to a hierarchical structure, the hierarchical organizational structure of the selected enterprise type category is traversed using the attributes of the energy consuming entity until reaching a leaf node, and 
 when the selected enterprise type category does not correspond to a hierarchical structure, the matching enterprise type category is selected; and 
   matching the energy consuming entity to an entity segment that corresponds to the selected enterprise type category or leaf-node, wherein the selected one of the trained machine learning models corresponds to the matched entity segment.   
     
     
         7 . The method of  claim 1 , wherein the generated forecast prediction comprises a daily energy usage forecast over a period of time, and the period of time comprises 1 day, 2 days, 7 days, 14 days, or 21 days. 
     
     
         8 . The method of  claim 7 , wherein the forecast prediction comprises at least a 90% accuracy with respect to an observed energy usage for the energy consuming entity over the period of time. 
     
     
         9 . The method of  claim 1 , wherein each trained machine learning model is trained using training data from energy consuming entities that match the trained machine learning model's entity segment. 
     
     
         10 . The method of  claim 1 , wherein the forecast prediction is used to perform energy usage planning for the energy consuming entity or energy grid planning. 
     
     
         11 . A non-transitory computer readable medium having instructions stored thereon that, when executed by a processor, cause the processor to generate time-series energy usage forecast predictions for energy consuming entities, wherein, when executed, the instructions cause the processor to:
 store a plurality of trained machine learning models, wherein each machine learning model is trained to forecast time-series energy usage data for one of a plurality of entity segments;   receive input data comprising historical time-series energy usage data for an energy consuming entity;   select one of the trained machine learning models for the energy consuming entity, wherein the selected trained machine learning model corresponds to an entity segment that matches the energy consuming entity; and   generate a forecast prediction using the selected trained machine learning model and the input data, wherein the forecast prediction comprises a time-series energy usage forecast for the energy consuming entity over a period of time.   
     
     
         12 . The computer readable medium of  claim 11 , wherein the entity segments are organized according to an organizational structure that comprises a plurality of enterprise type categories, and the organizational structure comprises hierarchical structures under at least a portion of the enterprise type categories. 
     
     
         13 . The computer readable medium of  claim 12 , wherein each entity segment corresponds to an enterprise type category or a leaf-node of the organizational structure. 
     
     
         14 . The computer readable medium of  claim 13 , wherein the hierarchical structures under each of the portion of the enterprise type categories are organized according to one or more enterprise parameters with respect to each enterprise type category. 
     
     
         15 . The computer readable medium of  claim 14  wherein the enterprise parameters comprise one or more of enterprise sub-types, enterprise size, or goods or services for sale. 
     
     
         16 . The computer readable medium of  claim 14 , wherein selecting one of the trained machine learning models for the energy consuming entity further comprises:
 selecting an enterprise type category that matches attributes of the energy consuming entity, wherein,
 when the selected enterprise type category corresponds to a hierarchical structure, the hierarchical organizational structure of the selected enterprise type category is traversed using the attributes of the energy consuming entity until reaching a leaf node, and 
 when the selected enterprise type category does not correspond to a hierarchical structure, the matching enterprise type category is selected; and 
   matching the energy consuming entity to an entity segment that corresponds to the selected enterprise type category or leaf-node, wherein the selected one of the trained machine learning models corresponds to the matched entity segment.   
     
     
         17 . The computer readable medium of  claim 11 , wherein the generated forecast prediction comprises a daily energy usage forecast over a period of time, and the period of time comprises 1 day, 2 days, 7 days, 14 days, or 21 days. 
     
     
         18 . The computer readable medium of  claim 17 , wherein the forecast prediction comprises at least a 90% accuracy with respect to an observed energy usage for the energy consuming entity over the period of time. 
     
     
         19 . The computer readable medium of  claim 11 , wherein each trained machine learning model is trained using training data from energy consuming entities that match the trained machine learning model's entity segment. 
     
     
         20 . A system for generate time-series energy usage forecast predictions for energy consuming entities, the system comprising:
 a processor; and   a memory storing instructions for execution by the processor, the instructions configuring the processor to:
 store a plurality of trained machine learning models, wherein each machine learning model is trained to forecast time-series energy usage data for one of a plurality of entity segments; 
 receive input data comprising historical time-series energy usage data for an energy consuming entity; 
 select one of the trained machine learning models for the energy consuming entity, wherein the selected trained machine learning model corresponds to an entity segment that matches the energy consuming entity; and 
 generate a forecast prediction using the selected trained machine learning model and the input data, wherein the forecast prediction comprises a time-series energy usage forecast for the energy consuming entity over a period of time.